Title: Modified logistic regression using the EM algorithm for reject inference

Authors: Billie Anderson; J. Michael Hardin

Addresses: Department of Mathematics, Bryant University, 1150 Douglas Pike, Smithfield, RI 02917, USA ' Culverhouse College of Commerce and Business Administration, The University of Alabama, Box 870223, Tuscaloosa, AL 35487, USA

Abstract: Reject inference is one of the key processes required to build relevant credit scorecard models. Reject inference is used to infer the good or bad loan status to credit applicants that were rejected by the financial institution. If rejected applicants data is not used in the updating of the credit scoring model, the model is biased because it will not be representative of the entire applicant population. Many reject inference techniques perform an extrapolation method to infer the good or bad loan status of the rejected applicants. The issues with extrapolation are discussed, and this study provides a novel reject inference technique in which the rejected applicants are included in the model estimation process. The extrapolation problem is avoided using the methodology in this paper. The newly proposed reject inference technique is shown to outperform the standard extrapolation technique using a simulation study.

Keywords: model accuracy; confusion matrix; credit scoring; extrapolation; logistic regression; reject inference; sensitivity; specificity; expectation maximisation algorithm; simulation; loan status.

DOI: 10.1504/IJDATS.2013.058582

International Journal of Data Analysis Techniques and Strategies, 2013 Vol.5 No.4, pp.359 - 373

Published online: 28 Feb 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article